AINEWS
PTC is the CAD and product design software behind much of the world's physical product engineering, and Schneider is paying $22.6B to add it to a power and data platform for the AI buildout.
Schneider Electric's $22.6 billion all-cash offer for PTC, announced Sunday, is the clearest signal yet that the AI datacenter buildout has stopped being a chip story and started being an industrial-software story. The $205-per-share bid, which values PTC's equity at roughly $22.6 billion and its enterprise value at $23.7 billion, is the third leg of a platform Schneider has been assembling since 2023. It positions the French power-equipment giant as the most ambitious bidder for the "physical AI" layer underneath the model labs.
PTC is not a chipmaker, and it is not an AI model lab. It sells the computer-aided design (CAD) and product-lifecycle management (PLM) software that engineers at Airbus, automotive suppliers, and industrial machinery makers use to draw, simulate, and manage physical products. Roughly 30,000 industrial customers rely on it. The company projected $2.69 billion to $2.75 billion in revenue and about $850 million in free cash flow for fiscal 2026, according to the joint press release. At 21 times 2027 adjusted EBITA, or 13 times with full synergies, per Constellation Research, Schneider is paying a control premium, with a 42.3 percent bump to the last close and 46.1 percent to the prior 30-day volume-weighted average.
The thesis is not about datacenters per se. It is about the rest of the industrial stack that datacenters are pulling with them.
Schneider's framing, what the companies call "Energy and Industrial Intelligence," fuses three pieces. The first is Schneider's own power management and grid hardware, the switchgear, UPS systems, and cooling kit that already sit inside most hyperscale facilities. The second is AVEVA, the British industrial-software firm Schneider bought in 2023 for roughly $10 billion, which models and operates physical plants. The third is Cognite, an industrial-data fabric Schneider closed on recently. PTC is the missing design layer: where AVEVA watches a factory run and Cognite streams its sensor data, PTC is where the factory was first drawn in software.
The deal paperwork promises €250 million in run-rate cost synergies by year three and roughly €800 million in revenue synergies over time, about $280 million and $900 million at the deal's stated exchange rate. The revenue claim, which values PTC's contribution higher than its standalone forecast, is the bet that the three platforms will sell to each other's customers. Schneider also said software will become about 24 percent of its total revenue after the deal closes.
The Register, which first moved on the story, framed it as "datacenter boom rains money on infra companies." That captures a piece of the truth: the AI capex cycle is dragging every adjacent industrial line into the same gravity well. But Schneider's own materials, and the Reuters reporting on the deal, point to a broader industrial-AI consolidation, not a narrow datacenter play.
The criticism lane is real. $22.6 billion is a large price for an industrial-software target, and integration risk is non-trivial: Schneider and PTC share thousands of industrial customers, and the platform promise depends on engineering teams agreeing on a unified data model. The thesis is exposed if datacenter capex cools, or if AI demand plateaus, leaving Schneider holding an industrial-software vendor it paid for at chip-cycle multiples. There is also a vendor-lock-in question for legacy manufacturing, aerospace, and automotive customers who may find their CAD, plant, and data tools bundled under a single owner.
The next test is execution, not announcement. Until the deal closes, the question for the industrial-AI crowd is whether Schneider can actually wire PTC's design tools to AVEVA's plant models and Cognite's data fabric in a way an Airbus engineer or a Toyota plant manager will pay to use. If it can, the $22.6 billion will look like the first industrial-AI platform deal of the cycle. If it cannot, it will look like a chip-cycle multiple paid for a CAD company.